Employing a bioinformatics approach to analyze prostate cancer gene expression profiles, we identified recurrent gene fusions/translocations in the majority of prostate cancers (Tomlins et al, Science 2005). Specifically, we identified the androgen regulatory elements of TMPRSS2 fused to the members of the ETS family of transcription factors including ERG, ETV1, ETV4 and ETV5. Analogous to hematological malignancies, gene fusions/translocations identified in prostate cancer may represent pathognomonic biomarkers and molecular sub-types of disease. In this application, we plan to focus our efforts on characterizing this new class of gene fusion biomarkers. Preliminary work done by our group and others suggest that molecular subtypes as well as transcript variants of gene fusions may be associated with clinical sub-types of prostate cancer. The central hypothesis of this application is that molecular sub-types based on gene fusions and variants will be useful predictors of the aggressive potential of clinically localized prostate cancer and thus guide treatment. Given this, we propose the following Aims: Specific Aim 1: Discovery and nomination of novel molecular sub-types of prostate cancer. Specific Aim 2: Characterize associations of molecular sub-types of prostate cancer with clinical outcome and/or aggressiveness of disease in a radical prostatectomy cohort. Specific Aim 3. Characterize associations of molecular sub-types of prostate cancer with clinical outcome and/or aggressiveness of disease using prostate needle biopsy samples.